Baidu Miaoda Upgrades Enterprise Version, Launches AI Coding Order Platform

The report cited two small companies with almost no technical staff that used Miaoda for core business. Liaosijin, a nearly 30-person niche clothing brand without an IT department or outsourcing budget, built a chain covering merchandise planning, supply chain follow-up, payments and inventory, reducing data errors. Fujian Lanhai Heishi, a new materials company without a dedicated R&D team, built the Quanxu Zhiguan office system and connected it to WeCom, allowing hundreds of employees to handle projects, travel approvals and email without switching windows. Both systems were built by founders and frontline business teams that had never written code, according to QbitAI.

The report said traditional digital transformation usually means buying standard SaaS and adapting business processes to the software, or commissioning custom development, which is costly and slow. Those paths were largely unavailable to the two small companies. Miaoda's third path lets people who know the business turn processes directly into systems.

For individual creators, Miaoda takes natural-language requirements and handles design, development and database setup, automatically assembling a runnable full-stack application that can be published as a web page, mini program or native app. Projects that once required product managers, UI designers and front-end and back-end engineers can now be completed by a single creator. The report cited AI director Ye Mao, who packaged his directing and visual workflow into the Movo AI image creation system, and tech blogger Feng Qiming, who built an ordering mini program for a friend who runs a barbecue stall.

Enterprise use is more complex. Applications must handle multi-person collaboration, role permissions, data security and long-term high availability. Much of the tacit knowledge that needs to be systematized exists only in frontline workers' heads, and traditional development requires translating it into documents and passing it through product and R&D layers. Miaoda's approach is to let frontline staff build and iterate systems in natural language while enterprises configure permissions and management foundations. General business can remain in fixed ERP systems, while fast-changing, highly customized edge and long-tail needs get a flexible development path.

Over the past year, Miaoda's users grew 200 times, it created 5 million commercially valuable applications, and nearly 200,000 people use it online daily, according to QbitAI. Of those creators, 87 percent could not write code. In the first half of 2026, Miaoda held a 33.4 percent market share by monthly active users, ranking first in China's AI-native no-code application generation platform industry, the report said.

The enterprise version focuses on how groups develop together and how applications run reliably over time. It introduces a three-level organization structure of enterprise, team and member. Administrators manage the upper level, member resources are independent, and enterprises can allocate Miaodian credits by team and manage personnel and usage quotas. Enterprises can also set fine-grained business roles and permissions, allowing people who know the business to join the building process while maintaining management boundaries.

Liaosijin's supply chain system illustrates this. After migrating to the enterprise version, the supply chain team focuses on material follow-up, garment scheduling and inbound QC; finance handles payment vouchers and two-way reconciliation with business documents; partner factories can only enter production progress in limited modules and cannot see core financial or full data. The team began validating requirements in May this year, migrated to the enterprise version and made more than 1,500 rounds of adjustments. The system links procurement and processing payments to business documents, updates inventory flows with business changes, and the team exports all data to Excel weekly for backup. No data loss has occurred so far, according to QbitAI.

For runtime assurance, the upgrade supports enterprise selection of backend resources by business scale and paid expansion of storage and concurrency limits on top of existing SLA service guarantees and backend keep-alive capabilities. Projects support shared backends and multiple environments, and single upload files are now capped at 500MB. Enterprises can download source code, bind custom domains or remove application watermarks. For projects requiring local deployment, the backend supports one-click deployment to local environments and one-click updates.

Fujian Lanhai Heishi used this foundation to make its custom system part of daily work. Its Quanxu Zhiguan system covers project management, travel approval and email processing and is mounted directly in WeCom. Hundreds of employees use it for project management and travel approvals without switching entry points, while the backend supports high-frequency concurrency, the report said. The company is in the new energy sector and has grown 300 percent annually for three years, creating many business process needs, but it has no dedicated R&D team. Its CIO and a student who had not yet graduated built the systems, with the student completing 90 percent of the work. They built two systems: one for office collaboration, project management, travel management and IT demand management, and another for operations management covering HR performance, safety training and weekly reports. The office collaboration system alone would have cost at least 500,000 yuan per year from a mature vendor, but it went live with Miaoda in one and a half months, saving that amount annually. A safety training management system with an outsourcing quote of 380,000 yuan was also built with Miaoda. The HR performance management system went from zero to live in seven days, compared with the fastest previous time of two months for similar systems. The project management module handles programs, sub-programs and projects, and the two systems use SSO.

At Baidu AI Day, the trend was framed as 'generative software.' Zhu Guangxiang, general manager of Baidu Miaoda, said future enterprise software will not have only one standard answer, and what companies really need is 'software that grows with the business,' which he described as the greatest value of generative applications.

Miaoda also launched its first AI Coding order platform. It connects enterprise demand-side users with professional creators in the Miaoda community and extends service from application development to supply-demand matching, transaction settlement and fulfillment guarantees. Enterprises can post new R&D needs or list existing Miaoda applications for professional takeover. Demand-side users can choose creators through the creator hall and personal pages, while creators can find matching projects in the demand square. After both sides agree, they sign on the platform, and funds are held in escrow by Alipay and frozen until project acceptance. Development remains on the platform. After a creator's first submission, the demand side can preview the dynamic application and check progress and results at any time; after acceptance, funds are automatically settled. Before acceptance, either party can request after-sales intervention in a dispute.

For enterprises, the platform adds a flexible R&D resource pool. Business teams can use natural language to build an initial version and then invite external experts to refine it, or post complete requirements and screen candidates through platform certification and portfolio review. Because app development, communication, result previews and fund milestones stay in one platform, enterprises can see real results early, the report said. For individual creators, passing portfolio review and merchant certification allows them to take orders and monetize legally. The platform provides business opportunities, communication channels and settlement infrastructure, but creators still need professional ability to understand the business, control quality and take responsibility for delivery. Miaoda remains an application development tool at its core; the order platform grows on its existing AI development capabilities and brings external creators into the enterprise application production and delivery chain.

The report also addressed the one-person company concept. Miaoda's goal is not that one sentence can make a company profitable automatically, but that one natural language instruction can start application development that once required product, UI, front-end and back-end roles. Enterprises and individuals can then mobilize publishing, operations and external creator resources to push a demo into real commercial circulation. The core value of a CEO is not writing every line of code; finding real needs, integrating resources, judging business priorities and taking responsibility for results remain necessary. AI compresses the threshold and startup cost of technical delivery, giving ordinary people small-team-level power and returning to frontline business staff the initiative to build tools directly. Real business systems cannot be conjured from one sentence, but a small team can continue shaping tools around its business approach. A sentence can be the starting point; running a business still depends on human judgment at key nodes.